Forecasting resort hotel tourism demand using deep learning techniques - A systematic literature review

Area of Science:

• Hospitality Management
• Data Science
• Artificial Intelligence

Background:

• Revenue management in hospitality relies heavily on accurate occupancy rate (OR) forecasting.
• The tourism sector's dynamic nature and online booking trends complicate traditional forecasting methods.
• Advanced technical skills and software are often required for precise OR prediction.

Purpose of the Study:

• To systematically review the application of Deep Learning (DL) techniques for occupancy rate prediction in the hospitality industry.
• To analyze trends in DL-based OR forecasting from 2017 to 2022.
• To answer research questions concerning variables, DL algorithms, and evaluation metrics in OR prediction.

Main Methods:

• A Systematic Literature Review (SLR) was conducted using the Snowballing methodology.
• Fifty relevant papers published between 2017 and 2022 were selected for analysis.
• Research questions focused on input variables, DL algorithms, and performance metrics.

Main Results:

• Five categories of variables were identified as influential in OR prediction.
• Long Short-Term Memory (LSTM) networks emerged as the most frequently used DL algorithm.
• Mean Absolute Percentage Error (MAPE) was the most common performance metric, with hybrid CNN-LSTM models showing promise.

Conclusions:

• Deep Learning offers advanced solutions for occupancy rate forecasting in hospitality.
• LSTM is a prevalent choice, but hybrid models like CNN-LSTM warrant further investigation for enhanced prediction accuracy.
• Continued research into hybrid DL models is recommended for optimizing revenue management strategies.